Fix direct image extraction dispatch

[jioffe502]

Summary

• Default GraphIngestor.extract() to extraction_mode="auto" so supported non-PDF inputs are routed through MultiTypeExtractOperator by extension.
• Preserve extraction_mode="pdf" as the explicit dedicated PDF/document graph path.
• Add coverage for BMP/TIFF/TIF page image materialization and mixed PDF+BMP auto dispatch.
• Add graph_ingestor.md documenting GraphIngestor usage, run modes, extraction modes, supported auto-dispatch groups, split configuration, and post-extraction stages.

Testing

• python -m pytest nemo_retriever/tests/test_ingest_interface.py -q
• Full JP20 pipeline smoke-tested with the auto dispatch path.

[greptile-apps]

Greptile Summary

This PR fixes auto-dispatch routing for direct image inputs (BMP, TIFF, TIF) by changing GraphIngestor.extract()'s default extraction_mode from "pdf" to "auto", so callers no longer need to pass extraction_mode="auto" explicitly for mixed or image-only corpora.

• graph_ingestor.py: one-line default change from "pdf" to "auto" in extract(), with updated docstring.
• test_ingest_interface.py: adds parametrized BMP/TIFF materialization tests and a mixed PDF+BMP auto-dispatch test, both using monkeypatch.setattr against private _MultiTypeExtractBase methods.
• graph_ingestor.md: new reference document covering run modes, extraction modes, the auto-dispatch extension table, split_config, and post-extraction stages.

Confidence Score: 5/5

Safe to merge; the change is a one-line default swap in extract() with accompanying tests and documentation, and the MultiTypeExtractOperator auto-dispatch path was already implemented before this PR.

The core change is small and well-scoped: extract() now routes callers through MultiTypeExtractOperator by default rather than requiring an explicit extraction_mode="auto". The underlying dispatch infrastructure predates this PR, so the risk of a broken path is low. The new tests exercise BMP/TIFF materialization and mixed-corpus dispatch, confirming the intended behavior works end-to-end with mocked inference stages.

No files require special attention beyond the minor _extraction_mode initializer inconsistency in graph_ingestor.py.

Important Files Changed

Filename	Overview
nemo_retriever/src/nemo_retriever/graph_ingestor.py	Changes extract() default from extraction_mode="pdf" to "auto" and updates the docstring; __init__ still initializes _extraction_mode to "pdf", creating a minor internal inconsistency.
nemo_retriever/tests/test_ingest_interface.py	Adds parametrized BMP/TIFF image materialization tests and a mixed PDF+BMP auto-dispatch test; both new tests mock private internals of _MultiTypeExtractBase rather than the public operator boundary.
nemo_retriever/src/nemo_retriever/graph_ingestor.md	New reference documentation covering GraphIngestor usage, extraction modes, auto-dispatch extension groups, split_config, and post-extraction stages.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["GraphIngestor.extract()"] -->|extraction_mode default| B{"extraction_mode"}
    B -->|'auto' NEW DEFAULT| C["MultiTypeExtractOperator"]
    B -->|'pdf' explicit| D["PDF/Document Graph"]
    B -->|'image' explicit| E["ImageLoadActor"]
    B -->|'text' explicit| F["TxtSplitActor"]
    B -->|'html' explicit| G["HtmlSplitActor"]
    B -->|'audio' explicit| H["MediaChunkActor → ASRActor"]

    C -->|".pdf .docx .pptx"| D
    C -->|".png .jpg .bmp .tiff .tif .svg"| E
    C -->|".txt"| F
    C -->|".html"| G
    C -->|".mp3 .wav"| H

    D --> I["_run_pdf_pipeline"]
    E --> J["_run_detection_pipeline"]
    I --> K["Final Result DataFrame"]
    J --> K

Loading

Prompt To Fix All With AI

Fix the following 1 code review issue. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 1
nemo_retriever/src/nemo_retriever/graph_ingestor.py:220
The instance attribute is still initialized to `"pdf"` in `__init__`, while `extract()` now defaults to `"auto"`. Anyone reading the class body first will conclude the natural default is PDF mode, when in practice `.extract()` (required for any ingestion) overrides it to `"auto"`. Aligning the two removes the misleading signal.

```suggestion
        self._extraction_mode: str = "auto"
```

_{Reviews (5): Last reviewed commit: "Merge branch 'main' into investigate/ima..." | Re-trigger Greptile}

[greptile-apps]

Mocking too deep in the stack — fragile tests ahead. Both new tests patch _MultiTypeExtractBase._run_detection_pipeline and _run_pdf_pipeline, which are private implementation methods two levels below the public boundary. The mocking-discipline rule says to mock at the boundary (the public operator entry point or the I/O seam), not inside the call stack. A future refactoring that renames or restructures these private helpers will silently pass these tests even though the real dispatch logic has changed.

Rule Used: Tests must mock all external service dependencies ... (source)

Prompt To Fix With AI

This is a comment left during a code review.
Path: nemo_retriever/tests/test_ingest_interface.py
Line: 92-95

Comment:
Mocking too deep in the stack — fragile tests ahead. Both new tests patch `_MultiTypeExtractBase._run_detection_pipeline` and `_run_pdf_pipeline`, which are private implementation methods two levels below the public boundary. The mocking-discipline rule says to mock at the boundary (the public operator entry point or the I/O seam), not inside the call stack. A future refactoring that renames or restructures these private helpers will silently pass these tests even though the real dispatch logic has changed.

**Rule Used:** Tests must mock all external service dependencies ... ([source](https://app.greptile.com/review/custom-context?memory=mock-external-services))

How can I resolve this? If you propose a fix, please make it concise.

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[jperez999]

make sure this is still up to date with all the changes we made recently.